Skip to content

Update VMs, CMake to 3.20.1, CUDA to 11.3, and pwsh to 7.1.3#17331

Merged
BillyONeal merged 20 commits intomicrosoft:masterfrom
BillyONeal:vm_update
Apr 26, 2021
Merged

Update VMs, CMake to 3.20.1, CUDA to 11.3, and pwsh to 7.1.3#17331
BillyONeal merged 20 commits intomicrosoft:masterfrom
BillyONeal:vm_update

Conversation

@BillyONeal
Copy link
Member

I tried updating to Ubuntu 20.04 as well but that's apparently not offered in a reasonable way on Azure right now.

@BillyONeal BillyONeal added the category:infrastructure Pertaining to the CI/Testing infrastrucutre label Apr 17, 2021
@NancyLi1013
Copy link
Contributor

NancyLi1013 commented Apr 19, 2021

The failures caused by cudnn depends on PR #17346.

For forest, related PR #16836.

@BillyONeal
Copy link
Member Author

@NancyLi1013 Thanks for pointing those out!

@BillyONeal
Copy link
Member Author

Looks like the many libvpx failures are #17287 ?

@BillyONeal BillyONeal force-pushed the vm_update branch 3 times, most recently from abb6546 to 1cf853e Compare April 20, 2021 07:59
@BillyONeal
Copy link
Member Author

/azp run

@azure-pipelines
Copy link

Azure Pipelines successfully started running 1 pipeline(s).

@cenit
Copy link
Contributor

cenit commented Apr 21, 2021

you removed cudnn auto-download for recognized cuda versions. Wasn't it accepted that it was ok in other PRs?

This is removing a feature otherwise which might be the base on which someone has built something... (me)

@BillyONeal
Copy link
Member Author

you removed cudnn auto-download for recognized cuda versions. Wasn't it accepted that it was ok in other PRs?

It was accepted in other PRs because from original contribution it looked like downloading any other closed source library, which would normally be OK. Unfortunately, it has been brought to our attention that this behavior is both (1) contrary to cudnn's license terms, and (2) contrary to anaconda.org's terms of service. As a result I feel forced to remove it. I did ask some nvidia folks to give us better options: https://twitter.com/MalwareMinigun/status/1384335333249097732

This is removing a feature otherwise which might be the base on which someone has built something... (me)

I'm sorry :(

@jacobkahn
Copy link
Contributor

I've removed this functionality from the nccl port as well in #17431. It would be great to get a CUDA 11.3-compatible NCCL version on the CI machines as well.

@BillyONeal BillyONeal force-pushed the vm_update branch 2 times, most recently from f91bfb7 to 1e87ec1 Compare April 22, 2021 03:37
@BillyONeal
Copy link
Member Author

/azp run

@azure-pipelines
Copy link

Azure Pipelines successfully started running 1 pipeline(s).

@PhoebeHui
Copy link
Contributor

PhoebeHui commented Apr 22, 2021

@BillyONeal, we noticed that many PRs failed with following failures on osx pipeline machine, could you please help take a look?

Unable to locate executable file: 'pwsh'. Please verify either the file path exists or the file can be found within a directory specified by the PATH environment variable. Also check the file mode to verify the file is executable.
/bin/bash --noprofile --norc /Users/vagrant/Data/work/_temp/749d1b35-0c54-49fd-8da4-9c21db16a076.sh
Traceback (most recent call last):
  File "/Volumes/data/work/1/s/scripts/file_script.py", line 34, in main
    gen_all_file_strings(path, getFiles(path), headers, output)
  File "/Volumes/data/work/1/s/scripts/file_script.py", line 9, in getFiles
    files = os.listdir(path)
FileNotFoundError: [Errno 2] No such file or directory: '/Users/vagrant/Data/installed/vcpkg/info/'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Volumes/data/work/1/s/scripts/file_script.py", line 41, in <module>
    main(sys.argv[1])
  File "/Volumes/data/work/1/s/scripts/file_script.py", line 37, in main
    except e:
NameError: name 'e' is not defined
##[error]Bash exited with code '1'.

@BillyONeal
Copy link
Member Author

@BillyONeal, we noticed that many PRs failed with following failures on osx pipeline machine, could you please help take a look?

Unfortunately I know very little about how the osx stuff works; Nicole was working on that

@BillyONeal
Copy link
Member Author

/azp run

@BillyONeal BillyONeal requested a review from NancyLi1013 April 25, 2021 07:22
… deploy. This resolves the deployment script often hanging with no feedback as to why.
(Perhaps VS2019 version 16.10 moved where these are placed? I've been defensive and left an attempt to find from the old location in place.)
…have CMake authoring errors rejected by 3.20.1.
…of disk space in CI and to respect --clean-after-build.
@BillyONeal
Copy link
Member Author

/azp run

@azure-pipelines
Copy link

Azure Pipelines successfully started running 1 pipeline(s).

@NancyLi1013
Copy link
Contributor

All checks have passed now.

@NancyLi1013 NancyLi1013 added the info:needs-maintainer-attention Lets the current 'on rotation' vcpkg maintainer know they need to look at this. label Apr 26, 2021
@@ -0,0 +1,125 @@
# Copyright (c) Microsoft Corporation.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should probably be a utilities.psm1, like in macOS, but not blocking

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can't shove .psm1s through Invoke-AzVmRunCommand though; we're taking this script and find and replacing # REPLACE WITH UTILITY-PREFIX.ps1 with it.

@BillyONeal BillyONeal merged commit a9b27ed into microsoft:master Apr 26, 2021
@BillyONeal BillyONeal deleted the vm_update branch April 26, 2021 17:27
@JackBoosY JackBoosY added the category:tool-update The issue is with build tool or build script, which requires update or should be executed correctly label Apr 27, 2021
@JackBoosY
Copy link
Contributor

Great! Thanks Billy!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

category:infrastructure Pertaining to the CI/Testing infrastrucutre category:tool-update The issue is with build tool or build script, which requires update or should be executed correctly info:needs-maintainer-attention Lets the current 'on rotation' vcpkg maintainer know they need to look at this.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants